This article will respond to some of the criticisms levied at our analysis in our five-part series on Exact Sciences (EXAS). We stand by our analysis and believe we have published a thoroughly researched and validated short thesis on Exact Sciences (the "company"). We believe the evidence of Cologuard's impending failure is overwhelming and believe it is the best short in our portfolio today. Our price target remains $3 per share based on our expectation that Cologuard will fail in at least one (and likely all) of the four areas needed for success: FDA approval, payer reimbursement, doctor prescriptions, and patient compliance. As a reminder if any one of these four areas fails, the company's commercialization efforts will fail.
We are first going to tackle a few the critiques found in Adam Feuerstein's (AF) article here:
AF Critique 1 - "'Alpha Exposure' accuses Exact Sciences of misleading investors by setting up an easy and rigged comparison between Cologuard sensitivity and specificity and historical data on FIT sensitivity and specificity. The fund points to Exact Sciences' recent investor slides comparing Cologuard to older, previously published FIT data as proof the company is misrepresenting the accuracy of its screening test. 'Alpha Exposure' is wrong. In the pivotal 'DEEP-C' study conducted by Exact Sciences, all patients underwent screening for colon cancer using Cologuard and FIT."
Adam Feuerstein misconstrues our arguments. We presented exactly what the company published publicly. We even referenced this in article one here, where we stated "First, it appears that Exact administered a FIT in its DeepC trial in order to compare the two tests in an identical patient population. Exact clearly states a FIT was administered as a comparison here. Exact then chose not to disclose the data generated in its trial and chose to instead use an outdated 2005 FIT study done in Japan by Morikawa as the comparator test."
We never stated that the company omitted any FIT data. We said that it was odd that the company did not release the FIT results from the DeepC trial, not even the top line results, as is customary for publication of preliminary top line results. Instead, the company provided historical FIT data from a different study, which was the study with the worst FIT outcomes we have seen. The company has never been shy about telling investors positive information, so it makes us very suspicious that the company has chosen not to reveal the performance of the FIT. Imagine if a biotech company touted successful top line results of its cancer drug without releasing the results of the trial's control arm. Exact's actions should be viewed with similar skepticism.
AF Critique 2 - "Cancer sensitivity is really 75%, not 88%. Pre-cancer sensitivity is 27%, not 43%, as 'Alpha Exposure' claims."
We re-ran our analyses with a cancer sensitivity of 75%, a pre-cancer sensitivity of 27%, and a specificity of 97%. As you can see below, the InSure FIT still outperforms in every category:
To be honest, we are unsure why AF would bring this up. His figures actually support our case because his data is better than the numbers from the Morikawa study. We already proved that the FIT is superior to Cologuard after adjusting for the differences in testing frequency, even with the Morikawa numbers. If there is any other way critics would like us to present the numbers, please let us know. We have yet to find a way to present the cumulative efficacy of FIT vs. Cologuard that results in a win for Cologuard.
We will now address critiques made by Seeking Alpha Contributors (SAC):
SAC Critique 1 - There is no proof that the cumulative efficacy of the various tests is relevant.
Exact Sciences' CEO Kevin Conroy speaks over and over about the importance of cumulative results and taking multiple tests. There are many examples of this. Quotes from the 3Q 2012 transcript here, 4Q 2012 transcript here, and Citi Global Healthcare Conference transcript here prove this:
"I would like to drive home the point that point-in-time sensitivity is not the most important part of this test clinically. The impact of this test will be felt from repeated screening and the cumulative sensitivity, like the pap smear, of detecting pre-cancers, which take a long time to develop and turn into cancer."
"I, again, don't have any reason to believe that the sensitivity will fall to 85% from 98%, but I want to make sure that we're setting expectations and driving home the point that the real value of this test is in the cumulative sensitivity like the Pap smear, which is -- which detects on average about 47% of precancerous lesions for cervical cancer. But the real value is in repeat testing and that increases the sensitivity over time."
"This shows you the power of repeat testing. This is why the pap smear has basically eradicated cervical cancer in screen population because we have a long period of time to detect something and you have over 50% chance to detect it in any given time, repeat testing will find these precancerous polyps."
To further prove this point, as we mentioned in article five, CMS clearly states its opinion on the bottom of page 44/top of page 45 here. CMS believes that the assumption of the conditional independence of repeat FIT/FOBT screens is reasonable due to the fact that a particular lesion may be bleeding during one test but not another. This drives home the point that we have been making: the cumulative results of repeat testing are incredibly important for patients tested with FITs.
However, CMS then states that this assumption does not hold for DNA stool tests because regardless of whether or not a particular lesion is bleeding, its gene mutation is unlikely to change and the chances of finding cancerous and pre-cancerous lesions will therefore not increase from of repeated stool DNA tests. We can now see that cumulative sensitivity is less important for Cologuard than for FIT.
In summary, we believe it is appropriate to compare tests based on their recommended frequency. This means comparing one year of Cologuard results with three years of cumulative FIT results (or two years vs. six years, or three years vs. nine years, etc.). The results are clear: over a three year period Cologuard is less effective than FIT.
SAC Critique 2- Contributors agree with AF that we cherry picked the best possible data for the InSure FIT, which is unlikely to be the case, and we only performed our results over three years, not allowing the cumulative effect of the Cologuard test to be seen.
Below we present the results of re-running our analyses with a cancer sensitivity of 75%, a pre-cancer sensitivity of 27%, and a specificity of 97%, and performed the test over a nine year period. These numbers were also mentioned by "dejerzeus" and "Energy-Trader" in their comments on Seeking Alpha here. As you can see below, the InSure FIT is still dominant:
SAC Critique 3 - It is unfair to run the analysis without taking into account the actual compliance adjusted rates of catching cancerous and pre-cancerous polyps.
We believe that compliance rates are directly related to the simplicity of the test and inversely related to the amount of feces that a person is forced to handle. Therefore we strongly believe that compliance will be better for the InSure FIT than Cologuard. In fact, poor compliance rates were one of the main reasons that Exact's first two tests failed. Nevertheless, we ran our analyses assuming 100% compliance for both the InSure FIT and Cologuard, which in our opinion is unfair to the InSure FIT because compliance for FIT will likely be higher than for Cologuard.
Contributors have commented that it is less likely that a patient takes an additional test in the subsequent years following a negative result. We wanted to address this criticism. As was suggested in the comments, we performed a three year study of the Cologuard test and the InSure FIT using the compliance and accuracy numbers deemed fair in our critic's comments. We assumed compliance rates of 95.1%, 79.9%, and 66.2% for years 1, 2, and 3 respectively, and we used a cancer sensitivity of 75%, a pre-cancer sensitivity of 27%, and a specificity of 97% for the InSure FIT. As you can see below, the InSure FIT still outperforms in every category and for less money, even after being adjusted for declining compliance over time.
SAC Critique 4 - If the Cologuard test is so bad, why hasn't there been any insider selling?
Given the rate at which Exact Sciences is burning through cash, and the amount of cash that it is going to need to commercialize Cologuard, the company will undoubtedly need to raise more capital. This is much harder to do if it appears that management doesn't believe in the company and is selling stock.
SAC Critique 5 - As seen here, "defiance" stated that we should have included a disclaimer, and shouldn't have criticized Exact Sciences for doing so.
Rather than hiding behind a meaningless disclaimer in smaller, harder to see font at the bottom of the page, we came out and said, "Below is the expected performance of the Cologuard test in a 10,000 patient group, identical to what the company used in its DeepC test. This assumes that the test is given once every three years, as suggested by the company. Below it, we show the FIT being administered to those same patients every year for three years." We were open, honest, and forthright about all of our assumptions.
SAC Critique 6 - As seen in the comments here, "Ulysses Benjamin Dover" states that "Also possible this hedge fund is in cahoots with a distant competitor or a company that wants to buy Exact on the cheap."
SAC Critique 7 - It isn't fair to criticize the company for using a historic control because they have not released the actual FIT data from the study.
If Cologuard performed so well against the FIT in the same study, why would Exact Sciences only release data for Cologuard without releasing data for the FIT comparator? The data is clearly available. If the data wasn't available, how would the company know that the Cologuard outperformed? It is suspicious that the company would release one piece of data without the other. After all, this isn't a company that is shy about announcing positive news.
Further, if the DNA testing component of the test is so valuable, why didn't Exact release the number of true positives that were caught because of it rather than the FIT included with Cologuard? The company claims that this component is valuable. Why not release top-line data to support this claim?
SAC Critique 8 - As can be seen here, "Energy-Trader" stated that, "Meanwhile, Cologuard demonstrated 66% sensitivity for 2cm pre-cancerous polyps. Your beloved FIT comes nowhere near that kind of performance."
The first part of the critique is true. Cologuard did in fact demonstrate 66% sensitivity for 2cm pre-cancerous polyps. That said, we are still unable to find any study that tracks this data point, and have yet to find it as an endpoint for any study, FIT or otherwise, which signifies to us that this is a meaningless clinical endpoint. With that being said, we have nothing to compare this Cologuard data point to. Not surprisingly, Exact didn't provide this information for the FIT. If you have this data, please post it in the comments below.
SAC Critique 9 - FITs and FOBTs as part of Digital Rectal Exams (DREs) are not recommended by the American Cancer Society as a stand-alone test for colorectal cancer.
This is true. It is also true that the NIH appears to think that DREs are a valid way to collect stool for colorectal cancer testing. Frankly, we do not understand this criticism. We were open and honest about this point and even included a link to the American Cancer Society article in part three when we wrote, "DREs are also performed to collect stool for sampling, which allows the doctor to perform the FIT or FOBT, as seen here and here."
We stand by our beliefs that getting tested using FITs or FOBTs during DREs is far better than doing nothing, and that if a patient is unwilling to undergo these screens while at the doctor's office as part of an annual check-up, than he or she is very unlikely to administer a Cologuard test to him or herself at home.
In addition, "Energy-Trader" stated here, that FOBTs during DREs are only 9% sensitive to cancer, which "disproves" that DREs can be used. Given that he or she is talking about FOBTs and not FITs, this is using the wrong comparator. Further, this is not substantially different from the FOBT performance cited by the company on slide 18 of its June 2013 report here.
SAC Critique 10 - The detection rates for stages I, II, III, and IV were never analyzed.
While we do not have the data for the Cologuard test, the complete data for the InSure FIT is below and can also be found and explained in detail here:
As can be seen above, there appears to be no correlation between the stage of the cancer and the probability that Insure FIT will detect the cancer.
SAC Critique 11 - You are an anonymous author who is short Exact Sciences stock
Yes, you are correct.